TTrades' Fractal Model on five markets

The idea, from zero

Price cannot turn around without first printing a swing high or swing low. So TTrades numbers the candles around every swing and builds everything on one bet: once candle 2 has closed and proved the turn, candle 3 should expand away from it. You do not try to catch the turn. You let the wick form and trade the body.

“Fractal” means the same picture sits inside itself. A daily candle 2 contains a 4-hour pattern, and a 4-hour candle contains a 15-minute one. The model looks for a moment when all of them point the same way.

The words you need

The whole trade, on one chart

The rules as tested, written down and committed on 30 October 2026, before any test was run

What had to be pinned down, and what was left out

TTrades’ material leaves some decisions open and disagrees with itself on a few. Each became a fixed choice, and the ones that matter became their own test below: daily bias or none, a wider bias, candle 2 closures only, a strict point of interest, the Ideal Formation, equilibrium, candle 2’s colour, entry timing, a positional entry and a higher-timeframe target.

Not tested: comparing correlated markets (SMT and PSP, which need a second market in the same test), T-Spot, weekly bias, order blocks, reading consolidation against continuation, partial profits and runners. A discretionary trader uses these. So a “not this time” here says nothing about TTrades’ own trading.

The first 45 tests allowed all three closures together and never tested continuations on their own. These 30 were written down and committed before they were run, on 1 November 2026, when the first 45 results were already known. They add tests and change nothing about the first 45.

TTrades uses the word “continuation” in several ways, so there are two named sets rather than one blend. Both keep the main rule and change only which closure counts as the setup candle. Candle 4 only: the previous candle was a reversal closure, and this one holds its low and closes above its high. Candle 3 only: the previous candle swept and failed, and this one holds its low and closes above its body. TTrades says a candle 3 closure can signal “a reversal or continuation depending on context”, and a fixed rule cannot read that context. Neither set is the true continuation.

The promising one is worth reading against what surrounds it. Its last two test windows lost money, the same rule on the same pair is negative on three of the other four markets, and it was not rerun with trading costs, which matter more when stops are this tight. It would need fresh data to say more. The engine’s grade for it says the next step is a demo account; this page does not suggest anyone trade it.

For the 15 main-rule tests and the 30 continuation tests the setups were redrawn from the raw bars with no engine functions, and every signal had to match. For all of them, each trade’s fill, stop, target and exit were replayed bar by bar outside the backtester. Nobody approved sample trades by hand; that review is deferred.

Each test window ran with a 20-bar lead-in, so 77 signals a whole-period reading has were not seen in the windows, and 10 were seen only there. 9 of the main-rule trades come from those. They are left in as run: the results are what the engine did, and the change to any main-rule average is at most 0.04R per trade. The promising continuation test is unaffected: all its trades are the same either way.

Limitations

• Before trading costs. Spread, commission and slippage would push every average down, not up. A tight stop makes this matter more.
• 75 tests. Any single good-looking row can be luck; read each one against all 75.
• Out of sample is May 2023 – Oct 2026, not the whole data period.
• One reading of a discretionary model. SMT, PSP, T-Spot, weekly bias, order blocks and context reading are not in it.
• Almost all of TTrades’ chart examples could not be dated, so the rules were checked against just one of them on real bars.
• CFDs on one broker’s feed, not the futures TTrades trades. A candle 2 depends on whether one wick passes another, and small differences between feeds can change that.
• Each test window started with only 20 bars of history, so a rule that looks further back reads a little differently there: the first days lack an earlier higher candle, and a gap from before the window is not seen. Read over the whole period instead, the main-rule averages move by at most 0.04R per trade.
• A bar that touches both the stop and the target counts as a loss.
• A backtest. Not live trading, and not a funded account.